High-density Words Research Articles

The effect of mild dementia on speech perception in quiet and noise

Difficulty understanding speech, particularly in background noise, is a common complaint of older adults. In quiet, speech perception is often assumed to be relatively automatic. In background noise, however, additional cognitive abilities likely play a more substantial role in successful communication (Koeritzer et al., 2018; Rönnberg et al., 2013). We used a word identification task in quiet and two levels of speech-shaped noise ( + 6 and + 3 SNR), to examine the impact of mild dementia on speech perception. In addition to noise, the number of similar sounding words (i.e., number of neighbors) also influences spoken word recognition — words with high neighborhood density (many competitors) are more difficult to perceive compared to low-density words (Luce and Pisoni, 1998). We predicted that individuals with dementia would perform significantly worse in conditions with the highest cognitive demands (high-density words, least favorable SNR). Preliminary results revealed that identification scores were significantly lower for conditions with more background noise and that, even after controlling for hearing loss, individuals with dementia had significantly reduced performance in all conditions. These results suggest that after accounting for differences in hearing, individuals with mild dementia still struggle with speech understanding, even in the absence of background noise.

Orthographic neighborhood density modulates the size of transposed-letter priming effects

We used transposed-letter (TL) priming to test the lexical tuning hypothesis, which states that words from high-density orthographic neighborhoods have more precise orthographic codes than words from low-density neighborhoods. Replicating standard TL priming effects, target words elicited faster lexical decision responses and smaller amplitude N250s and N400s when preceded by TL primes (e.g., leomn-LEMON) compared with substitution primes (e.g., leuzn-LEMON) overall. We expected that if high-density words have more precise orthographic representations (i.e., with each letter assigned to a specific position), then they should be relatively less activated by TL primes and should give rise to smaller TL priming effects. In line with our prediction, N250 (but not N400 or behavioral) TL priming was significantly smaller for high-density words compared with low-density words over posterior sites. Such an interaction was not observed for pseudoword targets. Consistent with the lexical tuning hypothesis then, this pattern suggests that the nature of the orthographic code used to access lexical representations differs depending on the number of neighboring words in the lexicon. We conclude by discussing how lexical tuning could be implemented in current models of orthographic processing.

The Relation Between Vocabulary Knowledge and Phonological Awareness in Children With Cochlear Implants.

Purpose The purpose of this study was to evaluate the relation between lexical knowledge and phonological awareness performance of children with cochlear implants. Method Thirty children with cochlear implants (aged 5-7 years), 30 children with normal hearing matched for age, and 30 children with normal hearing matched for vocabulary size participated in the study. Children completed a vocabulary knowledge measure and three phonological awareness tasks with words that had high and low neighborhood density. Results Children with cochlear implants performed more poorly than their age-matched peers and similarly to their vocabulary-matched peers on phonological awareness tasks. When performance was analyzed according to the neighborhood density of the target word, children with cochlear implants and age-matched children performed better with high-density words. Across all groups, vocabulary size correlated significantly with phonological awareness performance. Conclusion Children with cochlear implants demonstrate delays in both vocabulary knowledge and phonological awareness performance, but children with cochlear implants appear to take advantage of lexical information similarly to their age-matched peers.

Density and Distinctiveness in Early Word Learning: Evidence From Neural Network Simulations.

High phonological neighborhood density has been associated with both advantages and disadvantages in early word learning. High density may support the formation and fine-tuning of new word sound memories-a process termed lexical configuration (e.g., Storkel, 2004). However, new high-density words are also more likely to be misunderstood as instances of known words, and may therefore fail to trigger the learning process (e.g., Swingley & Aslin, 2007). To examine these apparently contradictory effects, we trained an autoencoder neural network on 587,954 word tokens (5,497 types, including mono- and multisyllabic words of all grammatical classes) spoken by 279 caregivers to English-speaking children aged 18-24months. We then simulated a communicative development inventory administration and compared network performance to that of 2,292 children aged 18-24months. We argue that autoencoder performance illustrates concurrent density advantages and disadvantages, in contrast to prior behavioral and computational literature treating such effects independently. Low network error rates signal a configuration advantage for high-density words, while high network error rates signal a triggering advantage for low-density words. This interpretation is consistent with the application of autoencoders in academic research and industry, for simultaneous feature extraction (i.e., configuration) and anomaly detection (i.e., triggering). Autoencoder simulation therefore illustrates how apparently contradictory density and distinctiveness effects can emerge from a common learning mechanism. OPEN RESEARCH BADGES: This article has earned an Open Data badge for making publicly available the digitally-shareable data necessary to reproduce the reported results. The data is available at https://osf.io/2qk5j/. Learn more about the Open Practices badges from the Center for Open Science: https://osf.io/tvyxz/wiki.

Accuracy and variability in early spontaneous word production: The effects of age, frequency and neighbourhood density

High rates of error and variability in early word production may signal speech sound disorder. However, there is little consensus regarding the degree of error and variability that may be expected in the typical range. Relatedly, while variables including child age, word frequency and word phonological neighbourhood density are associated with variance in word production accuracy and variability, such effects remain under-examined in spontaneous speech. This study measured the accuracy and variability of 234,551 spontaneous word productions from five typically developing children in the Providence corpus (0;11–4;0). Using Bayesian regression, accuracy and variability rates were predicted by age, input frequency, phonological neighbourhood density, and interactions between these variables. Between 61% and 72% of word productions were both inaccurate and variable according to strict criteria. However loosening these criteria to accommodate production inconsistencies unlikely to be considered erroneous (e.g. the target /æləɡeɪtəɹ/ pronounced /ælɪɡeɪtəɹ/) reduced this figure to between 10% and 17%, with the majority of word productions then classed as accurate and stable (48%–58%). In addition, accuracy was higher and variability was lower in later months of sampling, and for high frequency words and high density words. The author discusses the implications of these results for future research and the differential diagnosis of speech sound disorder, and presents an explanatory account of findings emphasizing the development of oral-motor skills and increasingly detailed phonological word representations.

Relationship Between F2 Slope and Intelligibility in Parkinson's Disease: Lexical Effects and Listening Environment.

Purpose There is a complex relationship between speech production and intelligibility of speech. The current study sought to evaluate the interaction of the factors of lexical characteristics, listening environment, and the 2nd formant transition (F2 slope) on intelligibility of speakers with Parkinson's disease (PD). Method Twelve speakers with PD and 12 healthy controls read sentences that included words with the diphthongs /aɪ/, /ɔɪ/, and /aʊ/. The F2 slope of the diphthong transition was measured and averaged across the 3 diphthongs for each speaker. Young adult listeners transcribed the sentences to assess intelligibility of words with high and low word frequency and high and low neighborhood density in quiet and noisy listening conditions. The average F2 slope and intelligibility scores were entered into regression models to examine their relationship. Results F2 slope was positively related to intelligibility in speakers with PD in both listening conditions with a stronger relationship in noise than in quiet. There was no significant relationship between F2 slope and intelligibility of healthy speakers. In the quiet condition, F2 slope was only correlated with intelligibility in less-frequent words produced by the PD group. In the noise condition, F2 slope was related to intelligibility in high- and low-frequency words and high-density words in PD. Conclusions The relationship between F2 slope and intelligibility in PD was affected by lexical factors and listening conditions. F2 slope was more strongly related to intelligibility in noise than in quiet for speakers with PD. This relationship was absent in highly frequent words presented in quiet and those with fewer lexical neighbors.

Variation in /u/ fronting in the American Midwest.

Previous research has suggested that a greater degree of social indexing of gender, race, and regional background is produced in linguistic contexts that promote phonetic reduction. The goal of the current study was to explore this hypothesis through an examination of the realization of an ongoing sound change in the American Midwest-/u/ fronting-as a function of four linguistic factors that contribute to phonetic reduction: lexical frequency, phonological neighborhood density, discourse mention, and speaking style. The results revealed minimal effects of the linguistic factors on the degree of /u/ fronting among talkers with greater overall advancement in the /u/ fronting change-in-progress, suggesting that the process of /u/ fronting is nearing completion among some American Midwesterners. However, the results also revealed more /u/ fronting in plain laboratory speech than in clear laboratory speech and in low-frequency, low-density words than in low-frequency, high-density words among talkers with lower overall advancement in the /u/ fronting change-in-progress. The directions of these effects are consistent with the hypothesis that social indexing is greater in reduction-promoting contexts. Further, the relative sizes of these effects suggest that speaking style contributes more to variability in social indexing than lexical properties, such as frequency and neighborhood density.

Implementing Evidence-Based Practice: Selecting Treatment Words to Boost Phonological Learning.

Word selection has typically been thought of as an inactive ingredient in phonological treatment, but emerging evidence suggests that word selection is an active ingredient that can impact phonological learning. The goals of this tutorial are to (a) review the emerging single-subject evidence on the influence of word characteristics on phonological learning in clinical treatment, (b) outline hypotheses regarding the mechanism of action of word characteristics, and (c) provide resources to support clinicians incorporating word selection as an active ingredient in their approach to phonological treatment. Research demonstrating the influence of the word frequency, neighborhood density, age of acquisition, and lexicality of treatment stimuli on phonological learning is summarized. The mechanism of action for each characteristic is hypothesized. Methods from the research studies are used to create a free set of evidence-based treatment materials targeting most of the mid-8 and late-8 consonants. Clinicians have numerous evidence-based options to consider when selecting stimuli for phonological treatment including (a) high-frequency and high-density words, (b) low-frequency and high-density words, (c) high-frequency and mixed-density words, (d) low-frequency and late-acquired words, and (e) nonwords. Incorporating word characteristics into phonological treatment may boost phonological learning. http://hdl.handle.net/1808/24768.

Interactions among lexical and discourse characteristics in vowel production

Various factors are known to affect vowel production, including word frequency, neighborhood density, contextual predictability, mention in the discourse, and audience. This study explores interactions between all five of these factors on vowel duration and dispersion. Participants read paragraphs that contained target words which varied in predictability, frequency, and density. Each target word appeared twice in the paragraph. Participants read each paragraph twice: as if they were talking to a friend (“plain speech”) and as if they were talking to a hearing-impaired or non-native interlocutor (“clear speech”). Measures of vowel duration and dispersion were obtained. Results from the plain speech passages revealed that second mention and more predictable words were shorter than first mention and less predictable words, and that vowels in second mention and low density words were less peripheral than in first mention and high density words. Interactions between frequency and mention, and density and mention, were also observed, with second mention reduction only occurring in low density and low frequency words. We expect to observe additional effects of speech style, with clear speech vowels being longer and more disperse than plain speech vowels, and that these effects will interact with frequency, density, predictability, and mention.

Distributional analyses in auditory lexical decision: Neighborhood density and word-frequency effects

In the present article, the effects of phonological neighborhood density and word frequency in spoken word recognition were examined using distributional analyses of response latencies in auditory lexical decision. A density x frequency interaction was observed in mean latencies; frequency effects were larger for low-density words than for high-density words. Distributional analyses further revealed that for low-density words, frequency effects were reflected in both distributional shifting and skewing, whereas for high-density words, frequency effects were purely mediated by distributional skewing. The results suggest that word frequency plays a role in early auditory word recognition only when there is relatively little competition between similar-sounding words, and that frequency effects in high-density words reflect postlexical checking.

Identification of lexical–phonological networks in the superior temporal sulcus using functional magnetic resonance imaging

General agreement exists that dorsal aspects of the temporal lobe support the perception of speech but there is less agreement regarding the mapping between levels of speech processing and neural regions within the dorsal temporal lobe. The present experiment sought to identify temporal lobe regions that support one such level, namely, lexical-phonological representation/processing. To do this, we manipulated phonological neighborhood density, a variable that affects processing within lexical-phonological networks. In a functional magnetic resonance imaging experiment, 10 participants listened to blocks of either high-density or low-density words. High-density words produced significantly more activation in the posterior half of the superior temporal sulcus bilaterally, suggesting that these regions are involved in lexical-phonological processing networks.

Blocking by word frequency and neighborhood density in visual word recognition: A task-specific response criteria account

Effects of blocking words by frequency class (high vs. low) and neighborhood density (high vs. low) were examined in two experiments using progressive demasking and lexical decision tasks. The aim was to examine the predictions of a task-specific response criteria account of list-blocking effects. Distinct patterns of blocking effects were obtained in the two tasks. In the progressive demasking task, a pure-list disadvantage was obtained to low frequency-high density words, whereas high frequency-low density produced a trend toward a pure-list advantage. In lexical decision, high-frequency words showed a pure-list advantage that was strongest in high-density words, whereas low frequency-low density words produced a trend toward a pure-list disadvantage. A simulation study implementing task-specific response criteria within the framework of the multiple read-out model provided an accurate description of the blocking effects obtained in the experiments. It is argued that adjustments of task-specific response criteria determine changes in list-blocking effects across different tasks.